January 13, 2013 Abstract
This paper considers changes in market comovement of merging US firms. Comparing the expected to the actual post merger comovement, we find that the post merger beta exhibits excess comovement with the acquiring firm. This suggests that the firm’s comovement is at least partly determined by its investors. We find that excess co- movement is significantly greater in cash transactions, when target shareholders tender their entire stake, than in pure stock transactions. Additionally, we document that the excess comovement is greater when the target is included in the S&P 500 as a result of the merger.
∗Financial support from the National Centre of Competence in Research ”Financial Valuation and Risk Management” (NCCR FINRISK) is gratefully acknowledged. Insightful comments and suggestions were received from Kjell Nyborg and Per Str¨omberg.
†Swiss Finance Institute - University of Zurich. Mailing address: Department of Banking and Finance, University of Zurich, Plattenstrasse 14, CH-8032 Zurich, Switzerland, Phone: +41-44-634-2956, Email: [email protected].
1. Introduction
Classical asset pricing theory predicts that in a frictionless market the return required by investors depends on the comovement of the firm’s assets with the market. In an international context, there is evidence that the comovement changes significantly when the location of listing changes (Froot and Dabora, 1999, and Chan, Hameed and Lau, 2003) and when a company is acquired by a foreign firm (Brealey, Cooper and Kaplanis, 2010). These results suggest that stock comovement with the market is at least partly determined by the firm’s investors and that international markets are segmented.
In the average merger, the majority of the target shareholders’ stake is acquired and therefore the post merger shareholder base is predominantly comprised of the acquiring firm’s shareholders. Given this, and if the market comovement is affected by the firm’s investors, we expect the post merger market comovement to be shifted towards the acquiring firm. This paper examines US mergers and provides evidence that investors partially determine stock comovement by showing a significant shift in market comovement towards the acquiring firm. We estimate the pre merger comovement of the target and the acquirer and use these estimates to calculate an expected post merger comovement. The expected post merger comovement is then compared to the actual post merger comovement. When the acquirer exhibits larger comovement with the market than the target (the prediction is asymmetric depending on the relative riskiness of the target and the acquirer), we find that the expected post merger comovement is 1.09 while the actual post merger comovement with the market is 1.18. This represents an excess comovement with the acquiring firm of 8.26 percent. Additionally, the implied effect on the target’s market comovement is an increase in beta of 0.27 or 34 percent relative to the pre merger beta.
Given that investors affect market comovement, the degree of excess comovement should be increasing in the fraction of equity tendered by target shareholders. Therefore, cash
mergers (which imply that target shareholders do not retain any stake in the merged firm) should be associated with significantly greater excess comovement. For cash mergers, the difference between the actual and expected post merger beta is 0.20 (compared to 0.09 for the overall sample). In cash transactions, target comovement increases by 0.32 or 46.1 percent relative to the pre merger beta. In contrast, for 100 percent stock deals (when there is less exit), the excess comovement is statistically and economically insignificant.
Building on work by Vijh (1994), Barberis, Shleifer and Wurgler (2005) argue that there is a ”habitat” of investors that invest in S&P 500 stocks. This implies that the firm’s share- holders change as a result of inclusion into the S&P 500 and therefore the comovement with the S&P 500 increases. Given that there is a S&P 500 habitat, we expect the excess comove- ment towards a S&P 500 acquirer to be larger when a target firm is included into the index as a result of the merger. Our results support this conjecture and additionally we verify that our results are not driven by an index inclusion effect.
It is well documented that investors have a preference for particular firm characteristics like industry and geographic location.1
Therefore, target shareholders that have a preference for a particular industry are more likely to sell their shares as a result of a cross industry merger than an intra industry merger. We find some support for this, in inter industry merg- ers (when the acquirer has a larger beta than the target) the excess comovement towards the acquirer’s comovement is 11.71 percent while for intra industry mergers it is 3.77 percent and statistically insignificant. Similarly, there is no excess comovement in within state mergers while in intra state mergers the excess comovement is 9.03 percent.
The findings of this paper suggest that there is not only cross-border segmentation, but also segmentation along other dimensions such as index membership and geography. However, there are a number of possible alternative explanations for our results that we
1
Empirically it has been documented that investors prefer stocks in their geographic vicinity (Coval and Moskowitz, 1999 and Huberman, 2001). Additionally, it has been shown that shareholders exhibit a preference for stocks from industries that they have experience from (D¨oskeland and Hvide, 2010).
have to consider. First, on average mergers are associated with increases in leverage and this could potentially explain the excess comovement.2 However, for leverage changes to explain our results, it must be the case that leverage increases when the beta of acquirer is greater than the beta of the target and that leverage decreases when the beta of the acquirer is smaller than the target’s beta, since in the first case we have a higher than expected post merger beta and in the latter a lower than expected post merger beta. In fact, for transactions in which the beta of the acquirer is lower than the target’s, we find that leverage increases modestly. Additionally, we conduct multivariate sorts that illustrate that excess comovement is independent of the change in leverage. Finally, in our regression analysis, we find that the change in leverage is insignificantly related to the excess comovement and does not affect our results qualitatively.
Second, some mergers result in synergies which might transform the assets and therefore also the comovement of these assets with the market. However, for synergies to explain our results it must be the case that the synergy asset has a riskiness that is above that of the expected post merger beta when the acquirer has a higher beta than the target and vice versa when the beta of the target is greater than that of the acquirer. Further, it must be that the transformation of these is rather rapid since we measure the post merger beta over 100 weeks after completion. Finally, in regression analysis we verify that synergies are not driving our results.
Third, following completion it is possible that the riskiness of the assets of the target is transformed to become similar to the riskiness of the assets of the acquirer. However, this risk transformation needs to be rapid (see above). Additionally, it has to be greater in transactions that are associated with greater shareholder exit (e.g., cash deals). Furthermore, we consider the progression of the firm’s post merger comovement and do not find evidence of a gradual transformation of the riskiness of the firm’s assets.
2
Ghosh and Jain (2000) study leverage increases in mergers. They find that leverage increases by a modest 6.3% on average. We find similar leverage increases in our sample (see Figure 2).
Prior work has provided evidence of segmentation by examining both returns and mar- ket comovements.3 Concerning comovement, Chan, Hameed and Lau (2003) document a decrease in comovement with the Hong Kong and an increase in the comovement with the Singapore Stock Exchange following a change in listing from Hong Kong to Singapore. Most closely related to us, Brealey, Cooper and Kaplanis (2010) document cross-border segmen- tation by providing evidence of excess comovement in cross-border mergers. They find that following a merger, the comovement with the exchange where the acquiring firm is traded increases while the comovement with the exchange of the target company decreases. We build on their results by providing evidence of segmentation by considering mergers of US firms. Another paper that illustrates within-border excess comovement is Pirinsky and Wang (2006). They show that when firms change headquarters location they start comoving more with an index of firms in the geographic vicinity of their new headquarters.
The remainder of the paper is organized as follows. In Section 2, we describe our data sources, sample selection criteria and methodology. We start Section 3 by considering sorts illustrating the relation between the pre merger and post merger betas. We then verify these results using regression analysis. In Section 4 we show that our results are not driven by asset transformation and Section 5 concludes.
2. Data and Methodology
Our sample of mergers and acquisitions comes from the Securities Data Corporation (SDC). We only include transactions between firms listed on the NYSE, AMEX and Nasdaq. More-
3
In terms of return segmentation, early work showed how investment barriers imply return premiums. The barrier to investment can be investment restrictions (Black (1974) and Stulz (1981)) or lack of information (Merton (1987)). In terms of empirical evidence, Hong and Kacperzyk (2009) show that ”sin” stocks exhibit abnormal performance that cannot be attributed to traditional factors. Additionally, Sloan and Lehavy (2008) and Bodnaruk and ¨Ostberg (2009) show that firms with less recogntion (segmented firms in terms of investor awareness) have higher returns.
over, our sample covers the period from 1980 to 2008.4 We only consider completed transac- tions where the target and acquiring company are publicly traded. Additionally, we require the target and acquirer to be different firms (i.e., we exclude all repurchases). This gives us a total of 8, 411 mergers. We obtain stock return data from the CRSP daily files (this reduces our sample to 6, 160).
In estimating comovement (see next section), we follow Brealey et al. (2010) and require 100 weeks of return data for the target and the acquirer prior to the run-up period and for the merged company after completion. This leaves us with 3, 510 deals.
Further, we only consider deals where 100 percent of the target company is owned by the acquirer after the merger. We only include targets which have a market capitalization above 50 million (Hackbarth and Morellec, 2008). In order to evaluate if the post merger comovement is biased towards the acquirer, we require that the targets assets to represent a non-insignificant proportion of total assets of the merged company.5
Therefore, we only consider mergers in which the target company has a market capitalization that is at least 25 percent of the acquirer. Finally, we exclude deals which involve at least one financial firm (SIC code 6000 to 6999). This leaves us with a total of 712 deals.
To control for the change in leverage due to the merger we calculate the leverage change as defined by Ghosh and Jain (2000). Leverage is the fiscal year-end ratio of debt to total firm value. We measure debt as the book value of long-term debt (Compustat Item dltt) added to the debt in current liabilities (Compustat item dlc). Total firm value is the book value of debt added to the market value of equity. The change in leverage is defined as the difference in leverage between the fiscal year end before the announcement of the merger and the fiscal year end after the completion of the merger.
4
SDC includes transactions from before 1980 and after 2008, but these transactions are excluded due to other restrictions.
5
Brealey et al. (2010) do not have to implement such a restriction since they examine comovement with respect to different markets whereas we consider one market, but examine whether the acquirer determines a disproportionate share of the comovement.
We draw on Brealey et al. (2010) in calculating the synergies of the merger. Synergies are the market adjusted increase in market capitalization of the acquirer and target in the six weeks surrounding the announcement (three before and three after) as a percentage of the pooled firm.
Figure 1 describes the time line of our research design.
Insert Figure 1 here
We estimate the individual comovement of the acquirer, target and merged firm with the market (the value-weighted CRSP index) over the 100 week pre run-up period (acquirer and target) and over 100 weeks post completion (merged firm). To avoid confounding effects of news announcements and rumors, we exclude eight weeks prior (run-up) to the merger announcement (Schwert, 1996). This involves running the following weekly regression for the acquirer, target and merged company:
Rj,t = αj + βjRm,t+ εj,t
where j is a firm index, Rj.t is the return on the firm and Rm,t is the return on the CRSP value weighted index. To reduce the effect of outliers, we winsorize our betas at the one and 99 percent level.
We calculate the expected merged beta as:
E(β) = M VA M VA+ (1 − λ)M VT βA+ (1 − λ)M VT M VA+ (1 − λ)M VT βT (1)
where βA and βT are the pre merger comovements of the acquirer and target, respectively and M V refers to the market value of equity. If the acquiring firm has a significant toehold, the comovement of the target is already partly reflected in the comovement of the acquirer (Brealey et al., 2010). Put differently, if only a small stake is acquired in the target due to
the toehold then the comovement of the acquirer is not expected to change significantly. To control for this, equation (1) adjusts for the fraction of the target held by the acquirer at announcement (λ).
Table 1 presents descriptive statistics of our key variables.
Insert Table 1 here
On average, target companies are roughly half the size of acquiring companies. Target companies represent roughly 35 percent of total pre merger market capitalization. We can see that on average leverage increases from 23.91 percent pre merger to 31.74 percent post merger. Our descriptive statistics indicate that total synergies only represent a small fraction of the pre merger firm. Additionally, in most deals the acquirer does not have a toehold. The pre merger betas of the target and acquirer are similar and close to one. Turning to the expected beta (E(β)), as predicted, it is between the target and acquirer beta. Finally, the post merger beta (βM) is greater than the expected beta which is consistent with a leverage increase.
We use SDC to classify the following methods of payment: cash, stock, mixed and other. Dummy variables Cash, Stock, Mixed and Other take the value 1 if the deal is only financed with cash, only with stock, a mix of both and if other methods of payment are used.
3. Empirical Findings
3.1. Univariate Analysis
This paper tests whether the investors contribute to the comovement of the firm with the market. To do so we examine mergers and acquisitions. Given that target investors exit
following the merger, the post merger comovement of the firm should be closer to the co- movement of the acquirer than expected. Additionally, the greater the fraction of target shareholders that leave as a result of the merger, the closer the post merger comovement should be to the comovement of the acquirer.
In this section we provide univariate analysis of the relation between the expected and the actual merged beta. Our central hypothesis is that the comovement of the merged firm is closer to the comovement of the acquirer than expected. When the comovement of the acquirer with the market is greater than comovement of the target with the market (βA> βT), we expect the actual merged beta to be greater than the expected beta (βM > E(β)). Hence, implying that the acquiring firm exhibits undue influence (relative to its market capitalization) on the comovement of the merged firm. Likewise, we expect the actual merged beta to be lower than the expected beta (E(β) > βM) when the beta of the target is greater than the beta of the acquirer.
Figure 2 presents our pre merger betas (acquirer and target) and our post merger expected and observed beta. Panel A considers deals for which βA> βT while Panel B considers deals for which βT > βA.
Insert Figure 2 here
Examining Panel A, it is evident that the actual merged beta is greater than the expected beta indicating excess comovement with the acquirer. Turning to Panel B, we see that the actual merged beta is slightly below the expected beta.
Table 2 compares the actual to the expected betas in our overall sample, split according to whether βA is higher or lower than βT, and tests whether the excess comovement is significant.
When βA> βT, the expected beta is 1.09 compared to the actual merged beta of 1.18. The difference between the actual and expected merged beta (βM− E(β), excess comovement) is statistically significant at the one percent level and represents a shift towards the acquirer’s beta of 8.26 percent relative to the mean expected beta. This understates the effect on target betas since targets represent on average less than half of the merged firm.
To evaluate the economic impact on target betas, we calculate an implied target beta based on our estimates. We replace for E(β) in βM = E(β) by using equation (1) and rearrange to obtain an expression for the implied target beta,
βTImp = M VA+ (1 − λ)M VT (1 − λ)M VT b βM − M VA (1 − λ)M VT b βA (2)
Using our estimates bβA, bβM we calculate an implied target beta for each transaction. The implied target beta (βTImp = 1.05) is on average 34.2 percent larger than the pre merger estimated target beta (βT = 0.78) when βA> βT.
Turning to the deals in which βT > βA, we see that the excess comovement is negative (−0.01) which is in line with our prediction. However, the difference is not economically or statistically significant. One potential explanation for this finding is that in order to observe excess comovement we require that target investors sell their shares. Therefore, splitting our results according to method of payment (see next section) provides for a more powerful test.
3.1.1 Method of Payment
If equity comovement is determined by the firm’s investors, then the greater the fraction of target shareholders that exit following the merger, the greater the excess comovement (Brealey et al.) with the acquirer. In mergers that are paid only with cash, all target share- holders exit whereas in stock-for-stock mergers no target shareholder has to exit. Therefore, we expect the excess comovement with the acquirer to be significantly larger in cash mergers
than in stock mergers. Figure 3 presents pre merger and post merger betas of our cash deals according to whether βA > βT (Panel A) or βT > βA (Panel B).
Insert Figure 3 here
Both panels of Figure 3 are indicative of the post merger comovement having shifted significantly towards the comovement of the acquirer.
Panel A of Table 3 presents univariate analysis of pre and post merger betas of cash deals.
Insert Table 3 here
When the comovement of the acquirer is greater than the comovement of the target, the expected beta is 0.97 while the actual beta is 1.18 implying that the tilt towards the acquiring firm is 21.65 percent relative to the expected beta. Additionally, this difference is statistically significant at the one percent level. Further, the implied target beta calculated using equation (2) is 46.1 percent larger than the pre merger estimated target beta. Turning to the deals where βT > βA, we find an expected beta of 0.93 whereas the actual post merger beta is 0.84, the difference of −0.09 represents a −9.68 percent deviation from the expected merged beta. The implied target beta is now 27.5 percent lower than the pre merger beta. This difference is statistically significant at the 5 percent level.
Panel B of Table 3 presents our results for pure stock transactions. It is striking that irrespective of whether the target beta is higher or lower than the acquirer beta, the difference